Simultaneous achievement of energy-efficient operation and high thermal stability of magnetic devices by enhancement of Dzyaloshinskii-Moriya interaction and magnetic anisotropy energy

Energy-efficient operation and stable data retention are the key features of magnetic information devices. Simultaneous achievement of writing-energy reduction and data-stability enhancement has yet faced a dilemma, since both are subjected to the same governing mechanism of magnetization switching, and thus, it is not easy to reduce writing energy while keeping high thermal stability. Here, we propose a solution that bypasses the dilemma by introducing chiral spin alignment in magnetic structures, which assists the current-induced switching at pattern edges, with the energy barrier enhancement. Experiments on asymmetric Pt/Co/Cu/Pt films reveals that both the Dzyaloshinskii-Moriya interaction (DMI) and perpendicular magnetic anisotropy (PMA) are increased by inserting an ultrathin Cu layer. A large PMA enhances the thermal stability, whereas a large DMI reduces the switching current density by tilting the angle of chiral spin alignment at pattern edges. The present observation shows that an effective DMI engineering provides energy-efficient and highly stable magnetic structures suitable for spintronic applications.

Figure 1

(a) The easy-axis hysteresis loop measured by vibrating sample magnetometer. (b) The schematics for measuring the magnetic anisotropy field ${\mu }_0{H}_{\mathrm{K}}$ (upper part) and SOT efficiency $\varepsilon$ (bottom part), respectively. In the bottom part, the red dot denotes the laser spot for the MOKE signal and the yellow squares indicate the electrodes for current injection. (c) The normalized anomalous Hall effect signal $V_{\mathrm{H}}$ with respect to ${\mu }_0{H}_{\mathrm{y}}$. The symbols are measured data and solid lines are fitting lines. (d) Plots of $\varepsilon$ with respect to ${\mu }_0{H}_{\mathrm{x}}$. The vertical error bars are the standard deviation of several repeated measurements. The grey horizontal line indicates $\varepsilon =0$. The curved solid lines guide the eyes. All black, red, blue, dark green, and violet symbols and lines correspond to the samples with Cu 0, 0.1, 0.2, 0.3, and 0.4 nm, respectively.

Figure 2

(a) Schematic illustration of sample structure. A 6-µm-diameter Co/Cu/Pt circular dot is patterned on a 10-µm-width Ta/Pt cross Hall bar. (b)–(f) The switching probability $P$ with respect to field induction time $t$ for several different constant external fields ${\mu }_0{H}_{\mathrm{z}}$ as denoted inside each panel. The Cu thickness is also denoted in each panel. (g) The plot of $\ln \left(\tau \right)$ with respect to the $\left(1/{\mu }_0{H}_{\mathrm{z}}-1/{\mu }_0{H}_0\right)$. The symbols are measured data and solid lines are linear fitting lines. (h) The relative switching field with respect to the angle of applied external field. The magenta dashed line is the estimated trend from the Kondorsky model, and the green dashed line is from the Stoner-Wohlfarth model. In (g) and (h), all black, blue, red, dark green, and violet symbols and lines correspond to the samples with Cu 0, 0.1, 0.2, 0.3, and 0.4 nm, respectively.

Figure 3

(a) Current-induced magnetization switching loops with $H_z$ bias of $t_{\mathrm{Cu}}=0.2\mathrm{nm}$, measured by anomalous Hall voltage. As the larger $H_z$ was applied, $J_{\mathrm{S}}$ decreased. Plot of (b) $J_{\mathrm{S}}$ and (c) $E_{\mathrm{B}}$ with respect to $H_z/H_{\mathrm{C}}$ for different $t_{\mathrm{Cu}}$ as denoted in each panel. In (b), the vertical error bars are the standard deviation of several repeated measurements. In (c), the vertical error bars are the error of the $\mathrm{\Lambda }$ determination from the measurement of $\tau$.

Figure 4

(a) The MOKE images about the domain nucleation procedure at the edge. The image was taken at the 44-µm dot of $t_{\mathrm{Cu}}=0.2\mathrm{nm}$. The dot was initially saturated to the down domain (left part) and the up domain was nucleated by SOT due to chiral spin alignment at the edge (right part). The current pulse of amplitude $J=1.06\times {10}^{11}\mathrm{A}/{\mathrm{m}}^2$ with width of $30\mathrm{ms}$ was injected. The bright (dark) state corresponds to the up (down) domain. (b) Plot of $J_{\mathrm{S}}$ with respect to ${\mu }_0{H}_{\mathrm{DMI}}/{\mu }_0{H}_{\mathrm{K}}$ with several different ${\mu }_0{H}_{\mathrm{z}}/{\mu }_0{H}_{\mathrm{C}}$ as denoted in the panel. The samples ranging from the smallest ${\mu }_0{H}_{\mathrm{DMI}}/{\mu }_0{H}_{\mathrm{K}}$ to the largest ${\mu }_0{H}_{\mathrm{DMI}}/{\mu }_0{H}_{\mathrm{K}}$ indicate that with Cu 0, 0.1, 0.2, 0.3, and 0.4 nm, in ascending order. The vertical error bars are the standard deviation of several repeated measurements and the horizontal error bars are estimated from the chi fitting errors of ${\mu }_0{H}_{\mathrm{DMI}}$ and ${\mu }_0{H}_{\mathrm{K}}$. In (b), the vertical error bars are the error of the $\mathrm{\Lambda }$ determination from the measurement of $\tau$.